Pine A64+ Gigabit issue...
#11
(05-03-2016, 10:31 PM)rhkean Wrote: I installed the Ubuntu image from this link.

apt-get update while plugged into my gigabit switch:  reported 178b/s  (bytes per second)
I ctl-c'd it after 2 1/2 minutes on the 2nd source.
shutdown
plugged into 10mbps switch
ran the entire process under 30 seconds

Thx for ruling out other problem sources. We discussed in the meantime potential origins of this problem (most people with GbE boards aren't affected of) and thought about whether insufficient power supply might be the reason.

The external RTL8211E GbE PHY on the board needs a specific amount of power to work reliable and maybe connecting the board to a 100 Mbits/sec switch avoids the problem. Since still no one took the time to implement voltage readouts the only reliable way to check for a relationship with insufficient DC-IN is to ensure stable DC-IN, eg. powering through the Euler connector:

[Image: Pine64_Powered_through_Euler_Connector.jpg]

Another potential issue could be that there's something wrong with the 2 more cable pairs used for GbE. With Fast Ethernet only 4 wires/traces are used, with GbE all 8 are needed. But I would check for powering issues first.
  Reply
#12
Found it: Post 97: http://forum.armbian.com/index.php/topic...on/?p=8074

The PHY shouldn't need that much voltage (maybe even just 1.8V) but in case DC-IN voltage is insufficient strange things might happen.
  Reply
#13
(05-04-2016, 03:01 AM)tkaiser Wrote: Found it: Post 97: http://forum.armbian.com/index.php/topic...on/?p=8074

The PHY shouldn't need that much voltage (maybe even just 1.8V) but in case DC-IN voltage is insufficient strange things might happen.

a couple questions to clarify...
Are you saying to provide power from a controlled source via the Euler bus (pins 2, 4, 6, & 9) INSTEAD of over the micro-USB power port?

Also, I have a 3.5A uUSB wall wart that I bought for my Pi2 (I got tired of seeing the rainbow cube while watching videos in kodi). Currently, I'm just using the power supply that I bought with my backer kit. Do you think the 3.5A ps will help? Or, is the AXP803 pmic limiting the voltage (or current) through the board? (which is my guess, and thus powering via euler bus by-passes that... but, power management is not my strong suit)

thanks
rob
  Reply
#14
(05-04-2016, 06:38 AM)rhkean Wrote: Are you saying to provide power from a controlled source via the Euler bus (pins 2, 4, 6, & 9) INSTEAD of over the micro-USB power port?

Also, I have a 3.5A uUSB wall wart that I bought for my Pi2 (I got tired of seeing the rainbow cube while watching videos in kodi)

If the latter successfully prevented that RPi symptom then please give this PSU a try first (since then it's NOT a smart charger but provides the amount of amperage to Pine64 that is needed and not maxes out at 500mA).

Regarding Euler bus. When I made the picture above I simply got so sick of Micro USB related power problems that I used 4 cables. 2 are enough (5V to pin 4, GND to pin 6). And according to TL Lim and schematic the Micro USB based DC-IN and these Euler pins are connected directly (so when you power your Pine through the Euler pins you could also use the Micro USB barrel to charge phones and the like Wink

I would give the 3.5A PSU a try first (Pine64 won't exceed 1A with normal 'workloads' but the real problem are voltage drops mostly caused by USB cables with insufficient gauge ratings -- never try anything above AWG 20)
  Reply
#15
If you scale from 1000MBit to 10Mbit and everything works, doesn't Clock/Interrupt seem like a more likely cause of the problems people are seeing than power? If you change your CPU clock governor to performance and that affects your Ethernet performance, doesn't that suggest the Ethernet is on a governed clock? Are we using the right clock for the job? Is the clock configured correctly?

I will note on Gigabit settings an 80% packet loss on the local network at a packet rate of 1/s which is enough to reduce any TCP to a standstill.
  Reply
#16
(05-04-2016, 10:18 AM)Faye Wrote: I will note on Gigabit settings an 80% packet loss on the local network

And I have 0% packet loss.
  Reply
#17
Do you have the 2G REV B?


Sent from my iPad using Tapatalk
  Reply
#18
(05-04-2016, 06:32 PM)Faye Wrote: Do you have the 2G REV B?


Sent from my iPad using Tapatalk

I do, no issues. 

Code:
root@debianpine64:~# date
Thu May  5 02:18:24 UTC 2016
root@debianpine64:~# iperf -c 192.168.0.45
------------------------------------------------------------
Client connecting to 192.168.0.45, TCP port 5001
TCP window size: 22.5 KByte (default)
------------------------------------------------------------
[  3] local 192.168.0.164 port 47396 connected with 192.168.0.45 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  1.08 GBytes   929 Mbits/sec
root@debianpine64:~# iperf -c 192.168.0.45 -t 30
------------------------------------------------------------
Client connecting to 192.168.0.45, TCP port 5001
TCP window size: 22.5 KByte (default)
------------------------------------------------------------
[  3] local 192.168.0.164 port 47397 connected with 192.168.0.45 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-30.0 sec  3.25 GBytes   930 Mbits/sec
  Reply
#19
(05-04-2016, 07:00 AM)tkaiser Wrote: I would give the 3.5A PSU a try first (Pine64 won't exceed 1A with normal 'workloads' but the real problem are voltage drops mostly caused by USB cables with insufficient gauge ratings -- never try anything above AWG 20)

Well that didn't fix it.
This is the power supply:  http://www.amazon.com/Extra-Charger-HP-C...B00OPEW1I6

this is the what I did (on gigabit):
* new ubuntu image
* log in as ubuntu
* sudo su -
* apt-get update
* 20 minutes later, It's still on Hit 6 of 8.

this is the what I did (on 100mbps):
* new ubuntu image
* log in as ubuntu
* sudo su -
* apt-get update
* 8 hits, 15 seconds or so...


I'm not sure that I have any decent, non-usb wall warts lying around that I can use for power...
  Reply
#20
(05-04-2016, 09:13 PM)rhkean Wrote:
(05-04-2016, 07:00 AM)tkaiser Wrote: I would give the 3.5A PSU a try first (Pine64 won't exceed 1A with normal 'workloads' but the real problem are voltage drops mostly caused by USB cables with insufficient gauge ratings -- never try anything above AWG 20)

Well that didn't fix it.
This is the power supply:  http://www.amazon.com/Extra-Charger-HP-C...B00OPEW1I6

this is the what I did (on gigabit):
* new ubuntu image
* log in as ubuntu
* sudo su -
* apt-get update
* 20 minutes later, It's still on Hit 6 of 8.

this is the what I did (on 100mbps):
* new ubuntu image
* log in as ubuntu
* sudo su -
* apt-get update
* 8 hits, 15 seconds or so...


I'm not sure that I have any decent, non-usb wall warts lying around that I can use for power...

Have you run the kernel and uboot update scripts?
  Reply


Possibly Related Threads…
Thread Author Replies Views Last Post
  Pine A64+ mainline mipi-dsi Learnincurve 3 1,538 01-12-2021, 07:36 AM
Last Post: Learnincurve
  DietPi OS for PINE A64 MichaIng 0 709 12-15-2020, 06:11 AM
Last Post: MichaIng
  Fedora 32/CentOS 8 Pine A64+ Images wideawake 2 1,221 10-02-2020, 11:38 AM
Last Post: mathiraj
  LibreELEC(KODI) for Pine A64 (+) pineadmin 6 6,627 05-02-2020, 10:29 AM
Last Post: aaronmarsh632
  Lakka-nightlies for Pine A64 (and H64) roel 11 5,266 04-23-2020, 12:55 AM
Last Post: roel
  Read DHT22 on Pine A64 ayufan 3 2,680 01-19-2020, 02:51 PM
Last Post: {-DesT-}
  Centos 7 for Pine A64 Luke 2 2,145 05-28-2019, 12:18 AM
Last Post: pineadmin
  NEMS Linux for Pine A64 (+) Luke 1 1,906 05-09-2019, 05:42 PM
Last Post: pineadmin
  Pine Board using linux stuck during boot sequence ktaragorn 4 2,348 03-30-2019, 06:48 AM
Last Post: ktaragorn
  motionEyeOS (PINE A64(+)) pineadmin 4 6,346 11-09-2018, 03:25 AM
Last Post: pineadmin

Forum Jump:


Users browsing this thread: 1 Guest(s)